The impact of mixing conditions on the removal of water and solids from high-water, poor-quality bitumen froth was explored. Naphtha diluent and a demulsifier were added to improve removal of water and solids from bitumen froth. The mixing and subsequent settling of this system were carried out in the confined-impeller stirred tank, a lab-scale mixing test vessel with well-characterized, relatively uniform mixing conditions. A protocol for finding the proper demulsifier dosage at which to study mixing effects was applied successfully. High mixing energy J and the predilution of demulsifier (characterized by its injection concentration IC) improved dewatering and solids removal performance, agreeing with earlier studies in diluted bitumen and bitumen froth of higher quality (Laplante et al. Fuel Process. Technol. 2015, 138, 361–367; Arora, N. Mechanisms of Aggregation and Separation of Water and Solids from Bitumen Froth Using Cluster Size Distribution, 2016). An unexpected finding was that dewatering was significantly delayed in poor-quality froth: it was not detectable until up to 45 min in some cases. This induction time was replicated and was clearly impacted by changes in the mixing conditions.

The use of a focused beam reflectance measurement (FBRM) probe to track the chord length distribution dynamics in a bituminous system can be very challenging. Material coats the sapphire window of the probe and the quality of the data is compromised. The fouling index is a parameter used to assess the quality of the FBRM data. It is provided by the software and is the percentage similarity between two consecutive scans of the instrument, indicating how much of the reading has not changed due to material that is stuck to the optical window. When the FBRM was used in bitumen froth, the fouling index was higher than 90 %. In this work the significant reduction in fouling achieved with a sapphire repellency treatment is presented. The reduction in fouling was sufficient to detect broad trends in the FBRM data that will be used in future studies but was not sufficient to ensure quantitative measurement of the chord length distribution.

No abstract